This is the penultimate draft of a paper that has been published in Philosophia 39 (4): 741-750 [2011]. The final version is available here: http://dx.doi.org/10.1007/s11406-011-9316-7
Interests Contextualism Robin McKenna
University of Edinburgh
Abstract In this paper I develop a version of contextualism that I call interests contextualism. Interests contextualism is the view that the truth-conditions of knowledge ascribing and denying sentences are partly determined by the ascriber’s interests and purposes. It therefore stands in opposition to the usual view on which the truth-conditions are partly determined by the ascriber’s conversational context. I give an argument against one particular implementation of the usual view, differentiate interests contextualism from other prominent versions of contextualism and argue that, unlike those versions, interests contextualism can mitigate against the epistemic descent objection put forward by Duncan Pritchard in his 'Contextualism, Scepticism, and the Problem of Epistemic Descent' (the objection is that, on the contextualist view, an ascriber of knowledge cannot, for some subject S and proposition p, properly ascribe knowledge that p to S if that ascriber has previously retracted an earlier ascription of knowledge that p to S).
1. Introduction
In this paper I develop a version of epistemic contextualism (henceforth, contextualism) that I call interests contextualism. Interests contextualism is the view that the truth-conditions of ‘knowledge’ ascribing and denying sentences are partly determined by the ascriber’s interests and purposes.1 I argue that interests contextualism can, unlike other
versions of contextualism in the literature, mitigate against Duncan Pritchard’s epistemic descent objection (see Pritchard, 2001).
2. Contextualism and Relevance in a Context
Contextualism is the view that utterances of sentences involving the word 'knows' express different propositions relative to different contexts of utterance. Say that Dougal
1 In this paper I follow the practice of put quotation marks around ‘know’ and its
has the true belief that a certain bank is open on Saturdays. The contextualist thinks that Ted can truthfully utter the sentence ‘Dougal knows that the bank is open on Saturdays’ in his context of utterance but Jack can truthfully utter the sentence ‘Dougal doesn’t know that the bank is open on Saturdays’ in his (different) context of utterance.
Interests contextualism is a version of relevant alternatives contextualism (RAC):2
Relevant alternatives contextualism (RAC): 'S knows that p' is true in A's context C iff S's
evidence eliminates every alternative in which not-p that is relevant in C.
But what makes an alternative relevant? Let’s say that Ted is in the following practical situation. It’s Friday and he’s driving past the bank on his way home from lunch with Dougal and Jack when he remembers that he’s got a cheque that he can cash. He looks out of the window and notices there’s a long queue. He’s in no rush to cash the cheque so he could come back some other day. He decides to drive on, saying to himself ‘Dougal mentioned that he was in this bank last Saturday so he knows that it’s open then. I can come back tomorrow’. Call this Low. Jack is also driving past the same bank when he remembers that he’s got a big cheque that he has to cash by Saturday otherwise he’ll go bankrupt. He also notices the long queue but he decides to go in, saying to himself ‘Dougal said he was in this bank last Saturday, but banks do change their opening hours so he doesn’t know that’. Call this High.3 We can distinguish two options open to the
contextualist for determining relevance. First, it could be that salience makes an alternative relevant. In Low the alternative that the bank has changed its opening hours isn’t relevant because it isn’t even considered by Ted. In High that alternative is relevant because Jack takes it seriously. Call this the salience view. Second, it could be that stakes make an alternative relevant. In Low the alternative isn’t relevant because it isn’t important given Ted’s interests and purposes. In High it is relevant because it is important given Jack’s interests and purposes. Call this the stakes view. In this section I give reasons for thinking that the salience view should be rejected. An alternative that is salient in a context need not be relevant.
2 See Lewis, 1996 and Blome-Tillmann, 2009 for two other versions of RAC in the
literature.
Lewis, a defender of the salience view, provides the following rule governing what alternatives are relevant to an ascriber A of ‘knowledge’ that p to a subject S in context C:
Rule of Attention: Any alternatives in which not-p that are being attended to by A in C are
relevant in C (Lewis, 1996, p. 554).
For Lewis, what does it mean to attend to an alternative? Some have read Lewis as holding that the mentioning of an alternative suffices for it to be attended to (Blome-Tillmann, 2009, pp. 244-9). That would mean that any alternatives mentioned to or by an ascriber of ‘knowledge’ in her conversational context are relevant in that context. In his 'Knowledge and Presuppositions' Michael Blome-Tillmann argues for rejecting the
Rule of Attention by providing a case where our intuitions diverge from what the rule
predicts. A mother hears her teenage son slipping out of the house after midnight. After spending the night fuming, in the morning she challenges him: 'I know you went out late last night, I heard you!' to which he responds 'how do you know? It's possible that you dreamt the whole thing'. The son has mentioned an alternative on which it is not the case that he slipped out of the house. Blome-Tillmann says: 'On Lewis's account you find
yourself in a context in which you have to admit to your son that you do not 'know' that he sneaked away at night, and this surely is not just a pity, it is rather also mistaken" (op.cit. p. 246).
This case is a good counter-example to Blome-Tillmann’s reading of the Rule of Attention. Unfortunately, Lewis doesn't really think that just mentioning an alternative suffices to make it relevant. Let's say that, before her son gets up, the mother is discussing him with a friend. Her friend mentions the alternative that she dreamt her son leaving the house but only in jest. In such cases, says Lewis, "we might quickly strike a tacit agreement to speak
just as if we were ignoring [an alternative]; and after just a little of that, doubtless [that alternative] really would be ignored" (Lewis, 1996, p. 560). It isn't enough for the mother to be forced to
But would we want to say that the mother, in a case where the son sticks to his guns and insists that she might have dreamt the whole thing, couldn’t truly claim to ‘know’ that her son left the house? I certainly don’t want to say that, and I’m sure not many people would. I take it that our intuition is that if the mother were to say that she ‘knows’ then that would be true. It’s a real black mark against any rule of relevance that it delivers such a counter-intuitive result. So this counter-example gives us good reason to reject the Rule of Attention. It shows that an alternative that is being attended to in a context (it has been mentioned and the conversational participants haven’t implicitly agreed to ignore it) can be irrelevant.
3. Interests Contextualism
I have argued that one prominent version of the salience view should be rejected. In this section I develop a version of the stakes view. Consider the Low case from earlier. Ted is trying to decide whether to go to the bank on Friday or come back on Saturday and he has Dougal’s recollection as his evidence. This evidence is clearly not sufficient to rule out all alternatives in which the bank isn’t open. Banks can change their opening hours. But, even if that alternative is salient within Ted’s context (imagine that he has Tom in the car with him who repeatedly insists that banks can change their opening hours) it may not be that Ted should consider it. On my view, an alternative that an ascriber shouldn’t consider in a context is irrelevant. Compare this with Jack in High. Jack is in the same practical situation as Ted except that, for him, a lot rides on whether the bank is open on Saturdays. Even if the alternative that the bank has changed its opening hours isn't salient (imagine that, for whatever reason, Jack fails to consider it) it is relevant to Jack because he should consider it.
What I’m proposing is replacing the Rule of Attention with what I call the Rule of Relevance:
Rule of Relevance: If an ascriber should consider an alternative in which p in her context C
then that alternative is relevant in C.
it). The interests contextualist says that an alternative in which p is relevant to A in her context C iff A has a reason to consider the alternative in which p in C, and that what A has a reason to consider depends upon what the important practical questions are for A in C. In what follows I will clarify two points.4 First, what is it to have a reason to
consider an alternative? Second, what about cases where an ascriber mistakenly takes herself to be in a high (or low) stakes practical situation when she is actually in a low (or high) stakes situation?
Consider this case taken from Bernard Williams (Williams, 1980). Bernard has a glass in front of him and he's trying to decide whether to drink from it. He wants to drink from the glass if it contains gin, but he doesn't want to drink from it if it contains an unpleasant liquid such as petrol. Let’s say that the glass in fact contains petrol but Bernard is completely unaware of this. In fact, he believes that it contains gin. As Williams argues, in such a situation we would want to say that Bernard hasn’t got a reason to drink from the glass. Rather, he mistakenly thinks that he does (op.cit., pp. 78-9). Similarly, we would want to say that Bernard has got a reason to consider the possibility that the glass contains petrol but he mistakenly doesn’t think that he does. There isn't anything mysterious about these reasons. Bernard's practical interests and purposes (his desire to drink from the glass if it contains gin but not if it contains an unpleasant liquid) are such that he shouldn’t drink from the glass and he should consider the alternative that the glass contains petrol. It isn’t hard to make sense of reasons for agents to act in certain ways that exist in virtue of those agent’s motives and desires i.e.
internal reasons.
I understand what it is to have a reason to consider an alternative as follows: A has a reason to consider an alternative in which p in C iff A's practical interests and purposes in C are such that she should consider the alternative. When I say interests and purposes I mean the agent’s actual interests and purposes. If an agent mistakenly takes herself to be in a low stakes situation when she is actually in a high stakes situation, or vice versa, then that agent will have mistaken beliefs about what alternatives are relevant. But, on my view, that doesn’t affect what alternatives are actually relevant.
Consider Ted and Jack’s respective situations in Low and High. Take the alternative that the bank has changed its opening hours (and, in particular, that it is no longer open on Saturdays). Jack is an agent with certain practical interests and purposes. He has an important bill to pay, the bill won't be paid unless the cheque is cashed by Saturday, and he wants to ensure that the bill is paid. Given these facts about Jack, he has a reason to consider the alternative that the bank has changed its opening hours. His practical interests and purposes are such that he should consider the alternative. This is no more mysterious than the reason Bernard has to consider the alternative that the glass contains petrol in the case above. Compare this to Ted. He doesn't want to waste time standing in a queue when it isn't necessary, but otherwise it's all the same to him whether he goes to the bank on Friday, Saturday, or next week. Given these facts about Ted, Ted has no reason to consider the alternative. His interests and purposes are such that he need not consider the alternative. None of this would change if Jack were ignorant of his high stakes situation. He would still have a reason to consider the possibility that the bank has changed its opening hours. Similarly, if Ted were to take himself to be in a high stakes situation he would still lack a reason to consider that possibility.
Finally, one might wonder what interests contextualism has to say about a number of other cases discussed in the literature.5 I’ll consider two sorts of cases here. First, cases
where an ascriber is in a low standards context and discussing what a subject in a high standards context ‘knows’. Jason Stanley has argued that our intuitions about these sorts of Low Ascriber-High Subject cases are better handled by subject-sensitive invariantism than contextualism (Stanley, 2005, Chapter 6). Insofar as I defend a version of contextualism I hold that it is the ascriber’s context that matters to the determination of the truth-conditions of ‘knowledge’ ascriptions. I offer the following by way of response. I’m not arguing here for the superiority of interests contextualism over subject-sensitive invariantism. Doing that would require an in-depth consideration of what both views say about a wide range of cases. I’m arguing that there are problems with one prominent version of what I have called the salience view and, in response, putting forward a version of the stakes view that I call interests contextualism.
5 Thanks to an anonymous reviewer for Philosophia for pressing me on the need to clarify
Second, what about an agent in Jack's situation in High who doesn't care that he's going to go bankrupt if he doesn't cash his cheque? For this agent to not have a reason to consider the alternative that the bank has changed its opening hours it would have to be that his interests and purposes are such that he need not consider it. Again, when I say interests and purposes I mean the agent’s actual interests and purposes not what the agent believes his interests and purposes to be. If an agent truly lacks any interest or purpose that would be furthered by considering a certain alternative then, on my view, that agent has no reason to consider that alternative.
4. Interests Contextualism vs. Other Versions of Contextualism
In this section I highlight the main respects in which interests contextualism differs from some other versions of contextualism defended in the literature and discuss an objection to my view.
First, the differences. Say that A asserts 'S knows that p'. A's assertion is true in context C iff S's evidence eliminates every alternative in which not-p that is relevant in C. For the interests contextualist, standing in a certain relation to an ascriber's interests in a given context is sufficientfor an alternative in which not-p to be relevant in that context. Compare this to Lewis where what is being attended to within A’s context dictates which alternatives are relevant (Lewis, 1996, p. 559). Compare this to Keith DeRose where the course of the conversation within A's context dictates which alternatives are relevant (DeRose, 2009, Chapter 4).
Consider how the case of the mother and the insistent son from §2 might end. The mother, realising that her son is going to continue to insist that she might have dreamt him leaving the house, may put her foot down and claim ‘It doesn’t matter whether you insist that I might have dreamt it, I know that you left the house last night’. But, on Blome-Tillmann’s view, it’s ‘unclear’ whether that assertion is true (op.cit., p. 267).6 As
earlier, I take it that we have the intuition that the mother’s claim is clearly true. Unlike Blome-Tillmann, the interests contextualist can easily accommodate this. Given the mother’s interests and purposes, she hasn’t got a reason to consider the alternative that she dreamt her son leaving the house. She can truthfully assert that she ‘knows’ that he left the house.
I will now consider an objection to my view.7 In any conversation the participants will
have their own epistemic standards. On DeRose’s view, which he defends in Chapter 4 of his recent The Case for Contextualism, in a conversation between two conversational participants A and B where what he calls the personally indicated content of A and B differ, an ascription or denial of ‘knowledge’ to S is true/false iff S meets/fails to meet the standards set by the personally indicated content of both A and B, and truth-valueless iff S meets/fails to meet one set of standards but not the other (DeRose, 2009, pp. 144-5). What I have said above involves a denial of this view. On my view, it looks as if there can be one conversation with two separate standards. In response, I will argue that there are two objections to DeRose’s view.8
First, DeRose’s view has the consequence that, in conversational contexts in which the ‘knowledge’ ascriptions and denials of A and B differ in personally indicated content, if A ascribes or denies ‘knowledge’ that p to S then that ascription or denial has an indeterminate semantic content. As above, for DeRose, in such contexts ‘knowledge’ ascriptions or denials are truth-valueless. But the contextualist holds that the standards operative in a context play a role in determining the truth-values of ‘knowledge’ ascriptions by playing a role in determining the semantic content of those ascriptions. If
6 Thanks to an anonymous reviewer for Philosophia for pointing out that Blome-Tillmann
holds that the truth-value of an assertion in these sorts of contexts is unclear rather than, as I had claimed in an earlier version of this paper, false.
7 Thanks to an anonymous reviewer for Philosophia for pressing this objection.
8 I discuss both objections in a review of DeRose’s The Case for Contextualism that’s
a certain ‘knowledge’ ascription or denial is truth-valueless then, for the contextualist, that must be because it has an indeterminate semantic content.
Second, on DeRose’s view the standard contextualist response to certain skeptical problems isn’t available. Let’s say that Jim has just woken from an unusually vivid dream in which he loses his hands in an accident. Startled, he lifts his right arm, sees his hand, and affirms ‘I know that I have a hand, that was just a dream’. Later, a committed skeptic is informed of Jim’s utterance to which he response ‘Rubbish! Jim knows no such thing’. I take it that the standard contextualist response would be to say that both Jim and the skeptic spoke truly, and, if Jim were to have a similar nightmare in the future and, upon waking, affirm ‘I know that I have hands’ then that assertion would be true as well. But, on DeRose’s gap view, the contextualist can’t say this. DeRose discusses the following sort of case (op.cit., pp. 148-50). Say that in a low standards context Ted asserts that Dougal ‘knows’ that the bank is open on Saturdays. Later, Jack, who is in a high standards context and was told about Ted’s earlier assertion, says ‘Rubbish, Dougal knows no such thing!” DeRose says that Jack’s assertion is truth-valueless iff Dougal meets Ted’s standards but doesn’t meet Jack’s standards (op.cit., pp. 149-50). That’s exactly what’s happening, so Jack’s assertion is truth-valueless. That, of course, doesn’t change the fact that Ted’s assertion was true, or that what he asserted is still true in Jack’s high standards context. Now consider what happens when, later, Ted, still in a low standards context, reminds himself that Dougal ‘knows’ that the bank is open on Saturdays and asserts this. Presumably, on DeRose’s view this assertion is truth-valueless iff Dougal meets Ted’s standards but not Jack’s. That’s exactly what’s happening, so Ted’s assertion is truth-valueless. Ted’s earlier assertion was true, and still is true, but this later one isn’t. This case is structurally identical with the skeptical one above, so in that case DeRose would presumably have to say that, while Jim’s original assertion of ‘I know that I have hands’ was true, any later assertion of that same sentence would be truth-valueless. So the standard contextualist response to this sort of skeptical problem isn’t available to DeRose.
5. Epistemic Ascent and Descent
Say that Ted and Dougal are waiting for a flight to Toronto. They want to find out whether it has a stopover in Heathrow. Larry overhears the conversation and tells them that the flight itinerary says that it does. Ted asserts 'Larry knows that the flight has a stopover.' Call this context Low. Suppose that Dougal says 'hold on, we really have to meet Jack there, maybe the itinerary is inaccurate.' This, on some versions of contextualism, will lead to an upward shift in epistemic standards. Following Duncan Pritchard, call an upward shift in epistemic standards epistemic ascent and a downward shift
epistemic descent (Pritchard, 2001). In response, Ted says 'You're right. Larry doesn't know
that the flight has a stopover.' Call this context High. On all of Blome-Tillmann, DeRose and Lewis' versions of contextualism Ted's assertions in Low and High are both true.
A contextualist such as Lewis can explain an upward shift in epistemic standards, as apparently happens between Low and High. An alternative on which the itinerary is inaccurate isn’t salient in Low but is made salient in High and, as a result, the standards shift upwards. It’s rather unclear how a downward shift in epistemic standards occurs. DeRose suggests that once ‘the conversational air has cleared’ the standards will revert once more to the standards governing quotidian contexts (DeRose, 1995, p. 42). An explanation of how epistemic descent occurs would explain how, in a context, some time after Low and High, in which Ted and Dougal are no longer in a high stakes situation, Ted could truthfully re-ascribe ‘knowledge’ to Larry. Pritchard's objection is that even if the contextualist could provide an explanation of epistemic descent such that Ted could
truthfully assert Larry ‘knows’ that doesn't mean that Ted could properly assert that Larry
doesn't ‘know’ (Pritchard, 2001, pp. 336-41).9 Pritchard generates this objection by
appealing to what one conversationally implicates in ascribing knowledge to a subject. If I assert 'S knows that p' I imply, amongst other things, that S is an authority on the matter of p. If, later, I reverse my assertion and say ‘S doesn’t know that p’ I imply, amongst other things, that S isn’t actually an authority on the matter of p. So consider the implicatures generated by Ted's various assertions. In Low Ted implies that Larry is
9 This gives us another way of stating the second objection to DeRose’s gap-view in the
an authority on the matter of whether the flight has a stopover. In High Ted reverses this implicature and implies that Larry is actually no authority on this matter. If Ted were to later assert that Larry knows he would imply that Larry is an authority on the same matter after all. At this point, the mess of contradictory implicatures generated by Ted's various assertions leaves him unable to properly assert that Larry knows.
As Pritchard notes, this isn't meant to be a decisive objection against contextualism (op.cit., p. 341). There are two avenues of escape. First, the contextualist could accept that Ted couldn't properly assert that Larry ‘knows’. Second, the contextualist could deny that Ted couldn't properly assert that Larry ‘knows’. Taking the first option would mean accepting that once an ascriber has reversed an ascription of ‘knowledge’ to a subject based upon certain evidence then that ascriber could not then properly re-ascribe ‘knowledge’ to that subject based on that evidence. But consider what the contextualist wants to say about cases like Low and High. The contextualist idea is that in some contexts the epistemic standards are relatively low and so a large number of ‘knowledge’ ascriptions are true whereas in other contexts the standards are a good deal higher and so a large number of, if not all, ‘knowledge’ ascriptions are false. To take the first option is to hold that once the standards have got high they can’t then return to being low. Taking the second option would mean accepting that even though Ted is generating contradictory implicatures that is no obstacle to him making a proper assertion.10
Here’s what the interests contextualist can say in response. Pritchard exploits a volatile contextualist depiction of our epistemic discourse in which a shift in standards can occur at any time in any conversational context. This isn't the way that the interests contextualist depicts our epistemic discourse. For the interests contextualist, but not on the versions of contextualism defended by Blome-Tillmann, DeRose and Lewis, there is no shift in Ted's epistemic standards between Low and High. The possibility that the itinerary is inaccurate is one that Ted has a reason to consider in both Low and High: given his practical interest in meeting Jack, he should consider it. So Ted's assertion in
Low is false and his assertion in High is true. Of course, our practical interests do change
over time. Just because a given possibility isn't one that we have reason to consider in one context doesn't mean it isn't one we have reason to consider in another context.
10 Thanks to an anonymous reviewer for Philosophia for urging me to clarify Pritchard’s
Consider a modified airport scenario:
Low*: Ted and Dougal are waiting for a flight to Toronto and idly wondering how long it
will take. They want to find out whether it has a stopover in Heathrow. Larry overhears the conversation and tells them that the flight itinerary says that it does. Ted asserts 'Larry knows that the flight has a stopover.'
High*: On another occasion, Ted and Dougal are waiting for a flight to Toronto. They
need to stopover in Heathrow because it is imperative that they meet Jack there. Larry overhears the conversation and tells them that the flight itinerary says it does. Ted asserts 'Larry doesn't know that the flight has a stopover.'
We have specified that there is a change in Ted's practical interests between Low* and
High*. Say that, on some future occasion, Ted and Dougal are waiting for a flight to
Toronto and idly wondering how long it will take. Larry has again seen the itinerary and tells them that it is a stopover in Heathrow. It looks like Ted can’t properly assert that Larry ‘knows’ because doing so would generate a mess of contradictory implicatures. The two options above remain. Should the interests contextualist accept that Ted cannot properly assert that Larry ‘knows’ or deny that Ted cannot properly assert that Larry ‘knows’?
So the strategy I would urge in response to Pritchard's objection is to say that, in certain cases, such as Ted the frequent flyer, the interests contextualist should accept that an ascriber cannot properly make a ‘knowledge’ ascription that he regularly reverses. To do so would be to generate a mess of contradictory implicatures. The important point is that these will be isolated cases.
6. Conclusion
In this paper I have argued that what I call interests contextualism can at least mitigate against the force of the epistemic descent objection, unlike other prominent versions of contextualism in the literature. These other versions are views on which it is salience that determines the relevance of alternatives in a context. Because the interests contextualist thinks that stakes determine what alternatives are relevant in a context she doesn’t hold that our epistemic discourse is volatile.11
Bibliography
Blome-Tillmann, M, 2009, 'Knowledge and Presuppositions', Mind 118 (470), pp. 241-294
DeRose, K, 1995, ‘Solving the skeptical problem’, Philosophical Review, 104 (1), pp. 1-52 - 2009, The Case for Contextualism, Oxford: Clarendon
Lewis, D, 1996, 'Elusive knowledge', Australasian Journal of Philosophy, 74, pp. 649–667 Pritchard, D, 2001, 'Contextualism, Scepticism, and the Problem of Epistemic Descent',
Dialectica, 55, pp. 327-349
Stanley, J, 2005. Knowledge and practical interests, Oxford: OUP
Williams, B, 1980,‘Internal and External Reasons’, reprinted with postscript in Millgram, ed., 2001, Varieties of Practical Reasoning, MIT Press, 77–97
11 I would like to thank Jessica Brown for supervising the MLitt dissertation that this
